Research On Stable Economic Model Predictive Control And Its A Pplication In Power Generation

Model predictive control,as an advanced optimization control algorithm,uses receding horizon strategy to determine future control law by optimizing performance indexes.Model predictive control is widely used in complex industrial process control since it can directly deal with multi-variable and constrained problems,and it has become one of the most effective methods in current industrial control.The industrial process economy is traditionally handled in a hierarchical architecture,in which the upper layer realizes the economic optimization to obtain the set-point while the low layer realizes the set-point tracking using model predictive control.However,traditional control architecture tends to ignore the economic performance of the dynamic tracking process.Based on the control structure,the economic model predictive control strategy proposed in recent years integrates the economic optimization and tracking control into one layer,which means considering the economic performance of the system directly in the dynamic tracking process.And it realizes the efficient and economic operation of power production indeed.Economic model predictive control is a more advanced optimal control strategy,which is in line with the core concepts of intelligent manufacturing,market-driven manufacturing and 'Made in China 2025'.·At the same time,this economic model predictive control with arbitrary form of economic performance indexes as objective function also brings new challenges to its theoretical analysis.The power industry has undergone profound changes in the past decade.Improving energy utilization,saving costs and reducing equipment fatigue load are important tasks for power generation control system.Aiming at the thermal power generation and wind power generation control system,the feasibility,stability and robustness of economic model predictive control are studied in detail combining with fuzzy modeling and online optimization solution technology.Then an economic model predictive control strategy for actual electric power production is constructed.The main contributions are as follows.(1)The economic model predictive control strategy based on fuzzy modeling is constructed for boiler-turbine system.Due to the load dependent characteristic of the boiler-turbine system,a fuzzy model based on real-time operation condition is established to approximate the nonlinear characteristics of the thermal power system.With this fuzzy model,the linear feedback control law and stability region are designed to guarantee the recursive feasibility and stability.A finite time performance index is constructed to realize load tracking and improve the economy in the dynamic tracking process through receding horizon optimization.The simulation results under large load changing show the effectiveness of the proposed fuzzy economic model predictive control.(2)Fuzzy economic model predictive control strategy with time-delay compensation is designed for 1000MW ultra-supercritical unit.An augmented fuzzy model with embedded predictor is used to approximate the nonlinear time-delay model based on real-time operation condition.In order to design stable controller based on this augmented fuzzy model,the auxiliary controller and corresponding stability region are calculated by using the traditional fuzzy model predictive control strategy.Load tracking and economy improving of ultra-supercritical units are realized by receding horizon optimization of finite time performance indexes.The simulation results verify the effectiveness of this controller designed.(3)A nonlinear economic model predictive control strategy for 5MW wind power generation system is designed.The classical model predictive control obtains the optimal set-point in the upper layer according to wind speed information,and then achieves the control objectives by tracking the optimal set-point using predictive control in the lower layer.Different from the classical model predictive control.economic model predictive control directly uses the generated power as the objective function without calculating the optimal set-point according to wind speed.Maximum wind energy capture is achieved at high wind speed and rated power output is tracked at low wind speed.The economic model predictive control does not need to change the set-point in the objective function frequently,so it can effectively improve the utilization rate of wind energy,reduce system fatigue and prolong the service life of the unit when the wind speed varies around the rated value.(4)A robust economic model predictive control based on scenario-tree is investigated to solve the uncertainty of wind speed.The closed-loop economic performance and robustness of the closed-loop system are guaranteed by the explicitly scenario modeling of wind speed variables.Considering the control objective of reducing the fatigue load of tower,the Pareto optimal solution of the online multi-objective optimization problem is obtained by regulating the weight coefficients.The dynamic optimization strategy adopts simultaneous approach to reduce the computational burden,and the effective solution of large-scale dynamic optimization problems can be achieved.The simulation results show that the robust economic model predictive control can effectively improve the utilization rate of wind energy,meanwhile reducing the fatigue load of the system and maintaining the safe operation of equipment.